The Value of Follow-up Following Complete Remission with Frontline Chemotherapy for DLBCL

Introduction

Following completion of chemotherapy with curative intent for DLBCL, for those in remission (CR), numerous national and international guidelines suggest on-going follow-up primarily to detect relapse of disease. For example, the NCCN guidelines suggest that clinical history, examination and blood investigations should be performed every 3-6 months for 5 years and as clinically indicated thereafter. This is essentially a “screening programme” for an at-risk population that continues based on historical routine rather than due to evidence it is of benefit to the patient. Numerous papers have been published suggesting that routine surveillance scans or bloods such as LDH are not useful in this setting. We sought to analyse whether the screening procedures that are performed in the follow-up setting are useful in the detection of relapse in patients followed up in a single centre. Uniquely, we examined each time point separately rather than group all follow-up events together.

Methods

Data was obtained retrospectively from electronic patient records for all patients in remission following treatment with frontline chemotherapy for DLBCL at the Royal Marsden Hospital between 2005-2016. Screening assessments analysed were symptoms check, clinical examination, LDH >upper limit of normal, lymphocyte: monocyte ratio (LMR) <2.8, routine CT scans. Time-points for analysis of routine clinic appointments were derived from the British guidelines which advise 3 monthly follow up in the first year, 6 months in year 2 and annually up to 5 years. True positives were defined as patients who had a positive result and relapsed within that specified time period or within 6 months, whereas false negatives were only considered for those patients who had a negative test within the time period that they relapsed.

Results

Complete data from 262 patients who achieved a CR and were followed up was available. Median age was 63 years (17-94) with 59 patients with stage 1 disease, and 100 with stage 4 disease. The relapse rate was 59/261 (23%), with a median time to relapse of 13 months (2-120). Of the 59, 52 had symptoms they had noted prior to clinic, of whom 28 requested an early appointment to clinic and 22 waited until their routine clinic appointment or sought advice from other health professionals initially. Only 7 patients were picked up only on investigations performed at the hospital, 1 with hypercalcemia and 6 on routine scan with no symptoms, 3 at 3-6months, 2 and 9-12months and 1 at 12-18 months. Of these, 4/7 patients achieved a CR with subsequent therapy. The remaining 202 patients had 2853 clinic appointments, median number per patient 13 (1-44).

The negative predictive value of clinical history, examination, LDH, LMR and CT scan was consistently very high (>90%) at all time points. However, the positive predictive value was low for symptoms, 0-35% with the higher values being in the first year. Clinical examination had better PPV due to fewer positive results, including both true and false. LDH and LMR were associated with a poor PPV at all time points. PPV for CT scans was variable (Table 1).

Conclusions

The “screening” of patients who are in remission from DLBCL requires considerable health resources and cost and also leads to increased exposure of the patient to hospitals which during the COVID-19 pandemic is being discouraged. Likelihood of relapse (which decreases with time) has an impact on the effectiveness of a screening programme, the relapse rate in this cohort was 23% in keeping with other published data. Importantly, the majority of patients had symptoms suggestive of relapse and so the screening was not required. In the few that were asymptomatic when relapse was diagnosed, there was insufficient data to know whether this earlier detection led to a better outcome, however this has not been seen in other published cohorts. Finally the effectiveness of the screening is dependent on the predictive value of the “tests” being used and this dataset shows that whilst the negative predictive value is high, the positive predictive value is very variable and generally poor for all tools used.

We propose that educating the patient regarding symptoms of relapse and having patient directed clinical follow-up rather than routine appointments would lead to marked savings in health resources, reduce hospital exposure of patients by eliminating unnecessary visits without compromising the outcome of the patients.